Tech

Artificial Intelligence Can Now Accurately Describe Your Poop

A poop emoji plushie sitting on a table

We don’t have jetpacks, flying cars, or teleportation pods, as futurists of the ‘60s imagined we would, but artificial intelligence-assisted shit-scanning apps are amongst the many pleasures of the 21st century. One app, Dieta Mobile, uses AI to describe poop more accurately than patients themselves, according to a new randomized clinical trial presented at the Digestive Disease Week conference in San Diego over the weekend.

“Diseases of the gastrointestinal tract are extremely common. Irritable Bowel Syndrome (IBS) affects up to 1 billion people worldwide,” principal investigator Dr. Mark Pimentel told Motherboard. “Sometimes patients don’t even know they have problems until they realize what’s normal and what’s not normal.”

Videos by VICE

App users learn what’s normal by snapping photos of their poops before wiping. Since the only thing worse than a stranger accidentally seeing lewds in your camera roll might be them seeing a screen full of poop shots, photos are submitted directly in the app. Then artificial intelligence categorizes the poops into one of seven categories in accordance with the Bristol Stool Scale, a diagnostic stool tool. Normal poops appear “like a sausage with cracks on its surface” or “like a smooth, soft sausage or snake,” according to BSS.

Pimentel, who told Motherboard he had no financial stake in the app, said artificial intelligence takes the guesswork out of the diagnostic process, which generally relies on potentially inaccurate patient self-reporting. (According to the national transparency program Open Payments, Pimentel has served as a paid consultant for a pharmaceutical company that sells IBS drugs.) The study found that subjects appeared reluctant to use the full BSS, tending to gravitate toward the diarrhea (low) end of the scale. Investigators compared patients’ self-reported scores and AI-generated scores with “gold standards” generated by two expert gastroenterologists. Scores calculated by AI aligned more closely with the expert scores, compared with patients’ self-reported scores.

A version of the app is being promoted for clinician use. Curiously, a GPS feature allows doctors to track the aggressiveness of a patient’s (bowel) movements. “If a person has to get off the freeway and, you know, go to a convenience store or go to the bathroom, because they’re in desperate need of a bathroom,” said Pimentel, “the app will track that.”

High-tech poop projects entered the media spotlight in 2019, when researchers asked the public to #GiveAShit by sending photos of their poops to train artificial intelligence for the world’s first crowdsourced poop database. Since then a number of poop apps have entered the scene, including Moxie, a web-based app described by Wired as analyzing “turds with some precision,” but also recognized poop in the reviewer’s face.

“While practically tricky to use, this app could be a great way of getting in touch with our poo and its consistencies, for anyone who is not so au fait with the Bristol Stool Chart,” wrote Wired. “Taking seriously the decrees of an app that purports to tell me how healthy I am and recognises poo in my face? Perhaps not.”

These apps are part of a growing trend of AI-assisted healthcare. At its best, health-promoting technological design would be driven by the needs and desires of the most vulnerable patients, would be equally and freely accessible to all, and would allow communities to surpass gatekeeping inherent in the medical-industrial complex.

The pharmaceutical industry, however—ever more concerned with new gadgets and programs, rather than issues like poverty and pollution that are at the root of poor health— is pouring billions into AI research for the purposes of profiting, not ensuring people are healthy. Healthcare industries are “primed to profit from the potential of AI,” by harnessing “mass pools of data,” according to One Nucleus, a European group of pharmaceutical companies. Apps like Dieta Health are well-positioned to enter the AI-based market for clinical trials, which aims to boost pharmaceutical revenue by keeping research and development costs down.

For now, poop-scanning apps seem to be the least of our concerns, but it’s not difficult to imagine how seemingly objective algorithms could be utilized by the pharma industry to rubber-stamp drugs that will eventually be recalled from the market and the subject of lawsuits.